In the field of integrated circuit memory devices, there is a continuing trend toward memories that have store more information, consume less power, operate faster, take up less space, and cost less to make. While these are often competing interests, memory manufactures strive to make advances in these areas to remain competitive. Thus, the ability to manufacture small memory cells efficiently is crucial in maximizing the performance and cost-efficiency of a memory device.
Popular memories today include dynamic random access memories (DRAMs), static random access memories (SRAMs), read only memories (ROMs), and flash memories. Certain basic characteristics are shared by these memories. For example, these memories typically include one or more memory arrays, where each array has a plurality of memory cells arranged in rows and columns. Other than these basic characteristics, however, these memories possess many different attributes. By way of a general comparison, ROMs and flash memories do not exhibit true random access as do DRAMs and SRAMs. Also, DRAMs and SRAMS are volatile memories. DRAMS require constant power to retain and refresh the contents of the memory. SRAMs require constant power to retain the contents of the memory. ROMs, and flash memories are non-volatile memories. Furthermore, DRAMs typically require less area on a die than the other memories, but DRAMs generally do not exhibit the fastest access times. Thus, as can be appreciated due to the many trade-offs between these different memory configurations, the type of memory used greatly depends upon the requirements of the system in which it is used.
One reason for these differences may be understood by referring to the memory cells used by these various memories. Although the memory cells of these different memories store data in the form of an electrical charge, the memory cells take different forms. The form of a memory cell may dictate many of a memory's characteristics. For instance, the memory cell of a typical dynamic random access memory (DRAM) generally includes a memory element and an access device. The memory element is typically a small capacitor, which stores data as the presence or absence of an electrical charge on the capacitor. The access device, typically referred to as an access transistor, is electrically coupled to the small capacitor and controls the charging and discharging of the capacitor.
DRAMs possess many desirable features, such as large storage capacity, high storage density, and ease of manufacture. However, due to the type of memory cell used, DRAMs also require periodic refreshing, i.e., the capacitors need to be periodically recharged, to maintain the stored information. Although the memory cells of ROMs, and flash memories do not require refreshing, they suffer from disadvantages, such as lower storage densities, larger size, and greater cost to manufacture.
Instead of using memory cells that store information in the form of an electrical charge, memory cells may be manufactured of a material that is capable of storing information. Chalcogenides are a class of materials that may be used to store information in an integrated circuit memory. Chalcogenide material may be electrically stimulated to change states, from an amorphous state to increasingly crystalline states. In the amorphous state, chalcogenide material exhibits a high electrical resistivity. As chalcogenide material progresses into an increasingly crystalline state, its electrical resistivity generally decreases. Because chalcogenide material retains its programmed state even after removal of the electrical stimulus, chalcogenide-based memories are non-volatile. As an added benefit, chalcogenide elements may be repeatedly programmed into two or more states. Thus, chalcogenide-based memories may operate as traditional binary memories or as higher-based memories.
In chalcogenide-based memories, the memory cells are typically formed by disposing chalcogenide material between two electrodes. Examples of chalcogenide-based memories are discussed in U.S. Pat. No. 6,025,220 issued to Sandu; U.S. Pat. No. 6,087,689 issued to Reinberg; U.S. Pat. No. 6,117,720 issued to Harshfield; each assigned to Micron Technology, Inc. and each incorporated herein by reference. As discussed in these patents, U.S. Pat. No. 5,335,219 issued to Ovshinsky et al. provides an explanation of the function and operation of chalcogenide elements and their use in memory cells. U.S. Pat. No. 5,335,219 is also incorporated herein by reference.
A brief description of a conventional chalcogenide memory cell fabrication technique is now provided with reference to FIGS. 1A-1D. A bottom electrode 105 is formed on a substrate (not shown). FIG. 1A shows a chalcogenide, first material layer 107 formed on the bottom electrode 105. A second material layer 109 is formed on the first layer 107. The second layer 109 is then exposed to ultraviolet radiation, which drives the material of second layer 109 into the first layer. 107 to create a doped, active chalcogenide material layer 110 (FIG. 1B). A top electrode material is then sputtered on chalcogenide material layer 110 (FIG. 1C). The top electrode material may be a noble metal such as silver. However, the top electrode 115 includes agglomerations or protrusions 120 on the surface thereof (FIG. 1D). FIG. 2 shows a 5 μm×5 μm area atomic force microscopy image of the protrusions 120 on the upper surface of top electrode 115 with the chalcogenide material layer 110 being GeSe doped with Ag and the top electrode 115 being Ag. The top electrode is 1000 Å thick. The protrusions 120 have an average height of about 550 Å and create a surface roughness rns of 140 Å. The surface of the electrode also has a terrace area height around the protrusions of 95 Å. It is also noted that the protrusions are visible even for very thin electrode thicknesses such as 100 Å. In light of the foregoing, there is a need for fabrication of chalcogenide memory devices which reduce agglomeration of sputtered material on the chalcogenide.